Improve statistics error messages when sensor's unit is changing (#55436)

* Improve error messages when sensor's unit is changing

* Improve test coverage

homeassistant/components/sensor/recorder.py

@@ -200,11 +200,18 @@ def _normalize_states(
                     hass.data[WARN_UNSTABLE_UNIT] = set()
                 if entity_id not in hass.data[WARN_UNSTABLE_UNIT]:
                     hass.data[WARN_UNSTABLE_UNIT].add(entity_id)
+                    extra = ""
+                    if old_metadata := statistics.get_metadata(hass, entity_id):
+                        extra = (
+                            " and matches the unit of already compiled statistics "
+                            f"({old_metadata['unit_of_measurement']})"
+                        )
                     _LOGGER.warning(
-                        "The unit of %s is changing, got %s, generation of long term "
+                        "The unit of %s is changing, got multiple %s, generation of long term "
-                        "statistics will be suppressed unless the unit is stable",
+                        "statistics will be suppressed unless the unit is stable%s",
                         entity_id,
                         all_units,
+                        extra,
                     )
                 return None, []
             unit = fstates[0][1].attributes.get(ATTR_UNIT_OF_MEASUREMENT)
@@ -320,7 +327,7 @@ def compile_statistics(
                         entity_id,
                         unit,
                         old_metadata["unit_of_measurement"],
-                        unit,
+                        old_metadata["unit_of_measurement"],
                     )
                 continue
 

tests/components/sensor/test_recorder.py

@@ -1028,6 +1028,7 @@ def test_compile_hourly_statistics_changing_units_2(
     recorder.do_adhoc_statistics(period="hourly", start=zero + timedelta(minutes=30))
     wait_recording_done(hass)
     assert "The unit of sensor.test1 is changing" in caplog.text
+    assert "and matches the unit of already compiled statistics" not in caplog.text
     statistic_ids = list_statistic_ids(hass)
     assert statistic_ids == [
         {"statistic_id": "sensor.test1", "unit_of_measurement": "cats"}
@@ -1038,6 +1039,90 @@ def test_compile_hourly_statistics_changing_units_2(
     assert "Error while processing event StatisticsTask" not in caplog.text
 
 
+@pytest.mark.parametrize(
+    "device_class,unit,native_unit,mean,min,max",
+    [
+        (None, None, None, 16.440677, 10, 30),
+        (None, "%", "%", 16.440677, 10, 30),
+        ("battery", "%", "%", 16.440677, 10, 30),
+        ("battery", None, None, 16.440677, 10, 30),
+    ],
+)
+def test_compile_hourly_statistics_changing_units_3(
+    hass_recorder, caplog, device_class, unit, native_unit, mean, min, max
+):
+    """Test compiling hourly statistics where units change from one hour to the next."""
+    zero = dt_util.utcnow()
+    hass = hass_recorder()
+    recorder = hass.data[DATA_INSTANCE]
+    setup_component(hass, "sensor", {})
+    attributes = {
+        "device_class": device_class,
+        "state_class": "measurement",
+        "unit_of_measurement": unit,
+    }
+    four, states = record_states(hass, zero, "sensor.test1", attributes)
+    four, _states = record_states(
+        hass, zero + timedelta(hours=1), "sensor.test1", attributes
+    )
+    states["sensor.test1"] += _states["sensor.test1"]
+    attributes["unit_of_measurement"] = "cats"
+    four, _states = record_states(
+        hass, zero + timedelta(hours=2), "sensor.test1", attributes
+    )
+    states["sensor.test1"] += _states["sensor.test1"]
+    hist = history.get_significant_states(hass, zero, four)
+    assert dict(states) == dict(hist)
+
+    recorder.do_adhoc_statistics(period="hourly", start=zero)
+    wait_recording_done(hass)
+    assert "does not match the unit of already compiled" not in caplog.text
+    statistic_ids = list_statistic_ids(hass)
+    assert statistic_ids == [
+        {"statistic_id": "sensor.test1", "unit_of_measurement": native_unit}
+    ]
+    stats = statistics_during_period(hass, zero)
+    assert stats == {
+        "sensor.test1": [
+            {
+                "statistic_id": "sensor.test1",
+                "start": process_timestamp_to_utc_isoformat(zero),
+                "mean": approx(mean),
+                "min": approx(min),
+                "max": approx(max),
+                "last_reset": None,
+                "state": None,
+                "sum": None,
+            }
+        ]
+    }
+
+    recorder.do_adhoc_statistics(period="hourly", start=zero + timedelta(hours=2))
+    wait_recording_done(hass)
+    assert "The unit of sensor.test1 is changing" in caplog.text
+    assert f"matches the unit of already compiled statistics ({unit})" in caplog.text
+    statistic_ids = list_statistic_ids(hass)
+    assert statistic_ids == [
+        {"statistic_id": "sensor.test1", "unit_of_measurement": native_unit}
+    ]
+    stats = statistics_during_period(hass, zero)
+    assert stats == {
+        "sensor.test1": [
+            {
+                "statistic_id": "sensor.test1",
+                "start": process_timestamp_to_utc_isoformat(zero),
+                "mean": approx(mean),
+                "min": approx(min),
+                "max": approx(max),
+                "last_reset": None,
+                "state": None,
+                "sum": None,
+            }
+        ]
+    }
+    assert "Error while processing event StatisticsTask" not in caplog.text
+
+
 @pytest.mark.parametrize(
     "device_class,unit,native_unit,mean,min,max",
     [